I want to use the rule "compile-fail" to check the accuracy of some tests of the compiler, but there are some side effects that I don't know how to overcome.
Minimal example:
Jamfile.jam
import testing ;
compile-fail compile/1.cpp ;
1.cpp
not_the_code_to_be_compiled
I get the output
gcc.compile.c++ .build/test/1.test/gcc-4.7/release/1.o
test/compile/1.cpp:1:1: error: ‘not_the_code_to_be_compiled’ does not name a type
(failed-as-expected) .build/test/1.test/gcc-4.7/release/1.o
**passed** .build/test/1.test/gcc-4.7/release/1.test
...updated 3 targets...
The result suits me. Not satisfied with one small feature — the log is the diagnostic message of the compiler. Diagnostic messages can be large (e.g. when using templates) and I want they didn't get in the way. I would have made, if these messages were in text file or just displayed on the screen. I want my output looked something like this
gcc.compile.c++ .build/test/1.test/gcc-4.7/release/1.o
(failed-as-expected) .build/test/1.test/gcc-4.7/release/1.o
**passed** .build/test/1.test/gcc-4.7/release/1.test
...updated 3 targets...
Couldn't find in boost documentation. Maybe looking for the wrong thing or not there.